The Agent Skills Directory

[COMMAND_EXECUTION]: The skill constructs shell commands by interpolating user-provided input (the 'target' extracted from natural language questions) into a command template: python .claude/scripts/code_graph query <pattern> <target> --json. If the extracted target contains shell metacharacters and is not properly sanitized, it could lead to arbitrary command execution.
[EXTERNAL_DOWNLOADS]: The skill documentation lists dependencies such as tree-sitter, tree-sitter-language-pack, and networkx. These are well-known, legitimate libraries used for code analysis and graph management, and the skill assumes they are already installed in the environment rather than downloading them at runtime.

graph-query